Tuesday Aug 16, 2011

Browsers are ‘more than just a pretty face’ – a new look at client-side performance tuning

By Chris Papineau, JD Edwards Senior Principal Engineer

 I recently discovered what appears to be an excellent tool for collecting browser metrics while researching methods of JD Edwards performance analysis for the “client side”. By “client side”, of course, we no longer mean the “fat client”, but rather the so-called “thin client” or web-browser. Current JD Edwards trending methodologies do not measure any browser side metrics; they merely blast html traffic at a JAS server and measure the response times.

This, however, leaves an opportunity in the “thin client” part of the JD Edwards ecosystem. The browser includes the following very significant items which are arguably the cause of performance challenges just as often as server-side items.

1. Number of server roundtrips

2. Payload (size of data)

3. JScript runtimes

4. Network latency

5. Browser renderings

6. DOM related measurements

We can no longer think of the browser as merely some sort of “façade to a western town” which doesn’t do anything except look pretty. We cannot simply focus on the server side of interactive transactions, under the dated assumption that business functions are where the time is really spent, and the user interface layer is insignificant.

“Pay no attention to the man behind the curtain”

The browser has become less and less a “thin” part of the architecture as it may have been in the fledgling days of web browsers (NCSA Mosaic, anyone??). Today the browser is a complex piece of machinery in which business logic has crept its way in via JavaScript, where the rendering of large graphics is the norm, and very long grids spanning dozens of screens and hundreds of rows are commonplace.

In any case – the tool in question is called dynaTrace, and appears to be a one-stop shopping for answering the fundamental performance question on the web browser side: “Where is the code spending its time?” It is freeware, works on both IE and Firefox, is easy to use and install, has a very intuitive interface, and appears to provide data for all six critical performance metrics cited above. The premium version, which is available for a fee, is even capable of “Comparing Different Test Runs” per the claims on their website.

This appears to be right in the wheelhouse of what we need for client-side performance trending efforts. I will keep readers of this blog up to date on further developments. It may be a significant difference maker not only for client-side trending, but in our ability to resolve challenging performance problems more efficiently. Stay tuned…..

About

Welcome to the JD Edwards Blog. Post are contributed by the JD Edwards strategy and development teams, covering a broad range of topics. The goal of the blog is to create opportunities for information and idea sharing, to engage JD Edwards customers/users, partners, and Oracle employees in conversation, the spirit of improving the experience and effectiveness of all that participate in the community. Both EnterpriseOne & World products are fair game for blog topics. Topics will cover functional, technical, business topics,but it is not the intent of this blog to provide "support" activities and those discussions will be best served via My Oracle Support or Oracle Mix. We invite your comments and feedback, and look forward to lively conversation.

Search

Archives
« April 2014
SunMonTueWedThuFriSat
  
1
2
3
4
5
6
7
8
9
10
11
12
13
15
16
18
19
20
21
22
23
24
25
26
27
28
29
30
   
       
Today